caml-list - the Caml user's mailing list
 help / color / mirror / Atom feed
From: Virgile Prevosto <virgile.prevosto@m4x.org>
To: caml-list@yquem.inria.fr
Subject: Re: [Caml-list] Why type inference fails in this code
Date: Fri, 9 Oct 2009 10:22:18 +0200	[thread overview]
Message-ID: <20091009102218.639677c8@is010235> (raw)
In-Reply-To: <20091009070935.GA5141@kerneis.info>

Hello,

Le ven. 09 oct. 2009 09:09:35 CEST,
Gabriel Kerneis <kerneis@pps.jussieu.fr> a écrit :

> On Fri, Oct 09, 2009 at 07:25:41AM +0200, rouanvd@softwarerealisations.com wrote:
> > ======================================================
> > type t = MyInt of int | MyFloat of float | MyString of string ;;
> > 
> > let foo printerf = function
> >   | MyInt i -> printerf string_of_int i
> >   | MyFloat x -> printerf string_of_float x
> >   | MyString s -> printerf (fun x -> x) s
> > ;;
> > ======================================================
> 
> Type-inference has nothing to do with it, it's the type system itself
> which fails.
> 
> What is the type you would expect? I guess:
> (forall 'a. ('a -> string) -> 'a -> 'b) -> t -> 'b
> 
> This is impossible with Ocaml: you cannot have a universal quantifier,
> type variables are existantially quantified.
> 

Well, I'm afraid this the other way round: a polymorphic type in Ocaml
is implicitly universally quantified: (fun x->x : 'a -> 'a) means the
identity function can be applied to an argument of any type, not that
there exists a type 'a on which you can apply it.
In addition, the desired type for printerf is indeed an existential
one (printerf: exists 'a. 'a -> string -> 'a -> 'b): you don't want the
first argument to convert any type to a string, just the type of the
second argument.

A possible but very heavy solution would be to use functors and local
modules everywhere (I guess that Haskell's typeclasses would help here).

type t = MyInt of int | MyFloat of float | MyString of string ;;

module type Printerf=functor(A:sig type t val to_string: t -> string
end) -> sig val print: A.t -> unit
end

module Foo(Printerf:Printerf) = struct
let foo = function
| MyInt i ->
    let module M = Printerf(struct type t = int
                                   let to_string = string_of_int end)
    in M.print i
| MyFloat x ->
    let module M = Printerf(struct type t = float
                                   let to_string = string_of_float end)
    in M.print x
| MyString s ->
    let module M = Printerf(struct type t = string
                                   let to_string = fun x -> x end)
    in M.print s
end
;;

but personally I'd do the same thing as you:
> In you very specific case, though, I guess refactoring it is possible:
> let foo printerf = function
>   | MyInt i -> printerf (string_of_int i)
>   | MyFloat x -> printerf (string_of_float x)
>   | MyString s -> printerf s
> ;;

-- 
E tutto per oggi, a la prossima volta.
Virgile


  reply	other threads:[~2009-10-09  8:22 UTC|newest]

Thread overview: 7+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2009-10-09  5:25 rouanvd
2009-10-09  6:54 ` [Caml-list] " Vincent Aravantinos
2009-10-09  7:09 ` Gabriel Kerneis
2009-10-09  8:22   ` Virgile Prevosto [this message]
2009-10-09  8:41     ` Gabriel Kerneis
2009-10-09  8:59       ` Gabriel Kerneis
2009-10-09 10:22 ` Jeremy Yallop

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20091009102218.639677c8@is010235 \
    --to=virgile.prevosto@m4x.org \
    --cc=caml-list@yquem.inria.fr \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).